Explainable Machine Learning to Map the Impact of Weather and Soil on Wheat Yield and Revenue Across the Eastern Australian Grain Belt

Patrick Filippi (), Brett M. Whelan and Thomas F. A. Bishop
Additional contact information
Patrick Filippi: Precision Agriculture Laboratory, Sydney Institute of Agriculture, School of Life and Environmental Sciences, Faculty of Science, The University of Sydney, Sydney, NSW 2006, Australia
Brett M. Whelan: Precision Agriculture Laboratory, Sydney Institute of Agriculture, School of Life and Environmental Sciences, Faculty of Science, The University of Sydney, Sydney, NSW 2006, Australia
Thomas F. A. Bishop: Precision Agriculture Laboratory, Sydney Institute of Agriculture, School of Life and Environmental Sciences, Faculty of Science, The University of Sydney, Sydney, NSW 2006, Australia

Agriculture, 2024, vol. 14, issue 12, 1-20

Abstract: Understanding the causes of spatiotemporal variation in crop yields across large areas is important in closing yield gaps and producing more food for the growing global population. While there has been much focus on using data-driven models to predict crop yield, there is also an opportunity to use these empirical models to understand which factors are driving variations in yield and to quantify their contributions. This study uses a large database of 625 rainfed wheat yield maps from 14 different seasons (2007–2020) across the eastern grain belt of Australia. XGBoost models were used, with predictors including maps of soil attributes (e.g., pH and sodicity), along with weather indices (rainfall, frost, heat, growing degree days). The model and predictors could accurately predict field-scale yield, with a Lin’s concordance correlation coefficient (LCCC) of 0.78 with 10-fold cross-validation. SHapley Additive exPlanation (SHAP), a form of interpretive machine learning (IML), values were then used to assess the impact of the variables on yield. The SHAP values for each predictor were also mapped onto a grid of the study area for the 2020 season, which showed the impact of each predictor on wheat yield (t ha −1 ) and revenue (AUD ($) ha −1 ) in interpretable units. Weather variables, such as rainfall and heat events, had the largest impact on yield. Although generally less significant, soil constraints such as soil sodicity were still important in driving yield. The results also showed that despite their largely temporally stable nature, soil constraints impact yield differently, depending on seasonal conditions. Overall, data-driven models and IML proved valuable in understanding the impact of important weather and soil variables on wheat yield and revenue across the eastern Australian grain belt. This could be used to determine the magnitude and economic impact of soil constraints and extreme weather on crops across regions and to inform policies and farm management decisions.

Keywords: yield modelling; empirical; XAI; explainable AI; soil constraints; frost; heat (search for similar items in EconPapers)
JEL-codes: Q1 Q10 Q11 Q12 Q13 Q14 Q15 Q16 Q17 Q18 (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2077-0472/14/12/2318/pdf (application/pdf)
https://www.mdpi.com/2077-0472/14/12/2318/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jagris:v:14:y:2024:i:12:p:2318-:d:1545699

Access Statistics for this article

Agriculture is currently edited by Ms. Leda Xuan

More articles in Agriculture from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().